/nextflow-development

Troubleshooting

Quick fixes for common nf-core pipeline issues.

Contents

• Exit Codes
• HPC/Singularity Issues
• Pipeline Failures
• RNA-seq Specific
• Sarek Specific
• ATAC-seq Specific
• Resource Management
• Getting Help

Exit Codes

Common exit codes indicating resource issues (per nf-core docs):

Most pipelines auto-retry with 2x then 3x resources before failing.

HPC/Singularity Issues

Singularity cache issues

export NXF_SINGULARITY_CACHEDIR="$HOME/.singularity/cache"
mkdir -p $NXF_SINGULARITY_CACHEDIR

Using Singularity instead of Docker

On HPC systems without Docker, use Singularity:

nextflow run nf-core/<pipeline> -profile singularity ...

Note: For basic environment setup (Docker, Nextflow, Java installation), see the inline instructions in Step 1 of SKILL.md.

Pipeline Failures

Container pull failed

• Check network connectivity
• Try: -profile singularity instead of docker
• For offline: nf-core download <pipeline> -r <version>

"No such file" errors

• Use absolute paths in samplesheet
• Verify files exist: ls /path/to/file

Resume not working

# Check work directory exists
ls -la work/

# Force clean restart (loses cache)
rm -rf work/ .nextflow*
nextflow run nf-core/<pipeline> ...

RNA-seq Specific

STAR index fails

• Increase memory: --max_memory '64.GB'
• Or provide pre-built: --star_index /path/to/star/

Low alignment rate

• Verify genome matches species
• Check FastQC for adapter contamination
• Try different aligner: --aligner hisat2

Strandedness detection fails

• Specify explicitly: --strandedness reverse
• Common values: forward, reverse, unstranded

Sarek Specific

BQSR fails

• Check known sites for genome
• Skip for non-standard references: --skip_bqsr

Mutect2 no variants

• Verify tumor/normal pairing
• Check samplesheet status column: 0=normal, 1=tumor

Out of memory for WGS

--max_memory '128.GB' --max_cpus 16

DeepVariant GPU issues

• Ensure NVIDIA Docker runtime configured
• Or use CPU mode (slower)

ATAC-seq Specific

Low FRiP score

• Check library complexity in plotFingerprint/
• May indicate over-transposition

Few peaks called

• Lower threshold: --macs_qvalue 0.1
• Use broad peaks: --narrow_peak false

High duplicates

• Normal for low-input samples
• Pipeline removes by default
• Consider deeper sequencing

Resource Management

Set resource limits

--max_cpus 8 --max_memory '32.GB' --max_time '24.h'

Check available resources

# CPUs
nproc

# Memory
free -h

# Disk
df -h .

Getting Help

1. Check .nextflow.log for error details
2. Search nf-core Slack: https://nf-co.re/join
3. Open issue on GitHub: https://github.com/nf-core/<pipeline>/issues